Edmunds and Carpenter 01
Recovery of Diadema antillarum reduces macroalgal
cover and increases abundance of juvenile
corals on a Caribbean reef
Peter J. Edmunds*† and Robert C. Carpenter†
Department of Biology, California State University, Northridge, CA 91330-8303
Edited by Robert T. Paine, University of Washington, Seattle, WA, and approved February 6, 2001 (received for review November 2, 2000)
The transition of many Caribbean reefs from coral to macroalgal opinion that reefs in the Caribbean now are in a highly degraded
dominance has been a prominent issue in coral reef ecology for state (13).
more than 20 years. Alternative stable state theory predicts that The coral reefs along the north coast of Jamaica, particularly
these changes are reversible but, to date, there is little indication at Discovery Bay, provide the most familiar example of the
of this having occurred. Here we present evidence of the initiation collapse of Caribbean reefs. Discovery Bay has been studied
of such a reversal in Jamaica, where shallow reefs at five sites along extensively since the 1950s (1, 5, 14), arguably more than any
8 km of coastline now are characterized by a sea urchin-grazed other reef in the Caribbean, and has remained at the forefront
zone with a mean width of 60 m. In comparison to the seaward of the reports of a macroalgal phase shift along the north coast
algal zone, macroalgae are rare in the urchin zone, where the of Jamaica. Virtually identical trends have been reported at
density of Diadema antillarum is 10 times higher and the density of multiple sites in Jamaica (1), but the majority of these sites have
juvenile corals is up to 11 times higher. These densities are close to been monitored less frequently than those at Discovery Bay. In
the 1950s, the reefs at Discovery Bay were characterized by small
ECOLOGY
those recorded in the late 1970s and early 1980s and are in striking
contrast to the decade-long recruitment failure for both Diadema amounts of macroalgae, and scleractinian corals covered as
and scleractinians. If these trends continue and expand spatially, much as 90% of the substratum (14). By the 1990s, after two
reefs throughout the Caribbean may again become dominated by major hurricanes, decades of overfishing, the near-complete loss
corals and algal turf. of the keystone echinoid Diadema (8), increasing human popu-
lation pressure, and possible nutrification, Jamaican reefs were
dominated by macroalgae to a depth of 35 m, and coral cover was
M any coral reefs throughout the Western Atlantic region
have undergone dramatic changes in community structure
over the past two decades. The best known examples of these
reduced to less than 5% (1). However, in 1992 there were small
patches of Diadema on the forereef of Discovery Bay (15), and
changes are found in the Caribbean (1), where reefs that were by 1995–96, Diadema had become locally abundant (e.g., 1.8 per
formerly dominated by scleractinian corals and diminutive algal m2, ref. 16) in shallow water ( 6-m depth) and formed larger
turfs have become overgrown by macroalgae. This transition is patches of macroalgal-free substratum (P.J.E., unpublished ob-
referred to often as a phase shift to an alternate state (1–4), servations). Typically, these patches were scalloped-shaped, tens
where the alternative state (i.e., one dominated by macroalgae) of meters in width, and contained locally dense populations of
is considered stable unless, or until, a reversal of one or more of the herbivorous echinoids Diadema and Tripneustes ventricosus
the causative agents favors another change. However, there are (12, 16). Tripneustes normally is rare or absent in forereef
habitats (16), and its co-occurrence with Diadema in 1995–96
few examples of multiple phase shifts on coral reefs (5) and none,
suggested that it might play a pivotal role in initiating the
to our knowledge, of a reversal from macroalgal to scleractinian
removal of macroalgae (12, 16). By January 2000, the expansion
dominance. Here we present evidence from the north coast of
and coalescence of macroalgal-free areas formed contiguous
Jamaica suggesting that such a change has been initiated. On
zones hundreds of meters in length, suggestive of a reversal in
these reefs, there is localized recovery of populations of the sea
community structure. This study was designed to quantify the sea
urchin Diadema antillarum, a decrease in the abundance of
urchin zones on a larger scale and test the prediction that
macroalgae, and an increase in the abundance of juvenile corals.
recovery of Diadema populations both enhances coral recruit-
Although there are few comparable data spanning the last 30
ment and facilitates a return to coral dominance.
years in Jamaica, there is evidence that the present densities of
Diadema and juvenile corals have not been reported for more Methods
than a decade. If these patterns persist, the widespread recovery Study Sites. Five sites were selected haphazardly along an 8-km
of Diadema populations alone may result in reefs dominated section of the north coast of Jamaica (Fig. 1). At each site, the
once again by scleractinian corals and algal turfs. sea urchin and algal zones sampled were parallel to the shore and
Phase shifts on Caribbean coral reefs can be caused by a to one another and were between 4.5 and 8.5 m in depth.
variety of anthropogenic and natural factors (1, 6–8), and are However, within a site, the sampled portions of the sea urchin
integral features of communities that demonstrate multiple and algal zones differed by 2 m or less in depth. At most sites,
stable states (2, 3). However, their detection is a function of the Diadema were found in depths as shallow as 1 m, and algal zones
spatiotemporal scale of investigation (2, 9), and evidence of extended to 20 m in depth. Widths of the zone of highest
phase shifts has come only from reefs such as those in Jamaica
(1) and Hawaii (10), where decades of data are available. Even
such long-term studies can provide only equivocal evidence of This paper was submitted directly (Track II) to the PNAS office.
the underlying mechanisms of change; thus, there is still debate See commentary on page 4822.
concerning the role of bottom-up [i.e., nutrification (7)] vs. *To whom reprint requests should be addressed. E-mail: peter.edmunds@csun.edu.
top-down [i.e., herbivore (11, 12)] control in mediating macroal- †P.J.E. and R.C.C. contributed equally to this work.
gal phase shifts. Regardless of the mechanism, the numerous The publication costs of this article were defrayed in part by page charge payment. This
reports of macroalgal phase shifts together with the absence of article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C.
any documented reversals have contributed to the popular §1734 solely to indicate this fact.
www.pnas.org cgi doi 10.1073 pnas.071524598 PNAS April 24, 2001 vol. 98 no. 9 5067–5071
Fig. 2. Population densities of Diadema antillarum in sea urchin and algal
zones at five sites along the north coast of Jamaica. (Bars mean population
densities; error bars 1 SE; see Fig. 1 for site designations.) The size–frequency
distribution of Diadema pooled across all sites (sea urchin zones only) is shown
in the Inset graph with the percent occurrence of nine size classes based on the
Fig. 1. Map of Jamaica showing the location of the study sites along the maximum test diameter. Size-class designations represent the following
north coast. RB Rio Bueno, LTS Long Term Survey, M1 Mooring 1, DB ranges of maximum test diameters: 1 20 mm, 2 20 –29 mm, 3 30 –39 mm,
Dairy Bull, EDB East Dairy Bull ( 0.7 km east of DB), and * Discovery Bay 4 40 – 49 mm, 5 50 –59 mm, 6 60 – 69 mm, 7 70 –79 mm, 8 80 – 89 mm,
Marine Laboratory. Sites were selected to sample reefs that have been the and 9 90 –99 mm.
subject of long-term studies [RB, LTS, M1 (refs. 1, 11, 15, and 18)] and recent
surveys (DB; ref. 19), and to span the greatest scale accessible with small boats
(e.g., EDB). taken as the average of the two major diameters of the basal
portion of each colony. The most common genera at each
site zone were measured (Leptoseris, Porites, Siderastrea, Steph-
abundance of Diadema were measured perpendicularly from anocoenia, and Agaricia), by using colonies (n 50, except for
shore (EDB, DB) or from the reefcrest (M1, LTS, RB) to the Agaricia at DB, where n 100) that were selected haphazardly
point where benthic community structure changed abruptly from while swimming along the transect line used for counting
algal turf dominance (i.e., Diadema-grazed, ref. 17) to macroal- juvenile corals and community structure.
gae (measurements at two randomly selected locations at each
site). All surveys were completed during January 2000. Statistical Analyses. Echinoid abundance, community structure
(i.e., percentage cover), density of juvenile corals, and the
Echinoid Abundances and Sizes. Abundances of sea urchins were number of genera of juvenile corals per quadrat were compared
estimated in 1-m2 quadrats (n 20 per zone) that were randomly among sites and between zones with a two-way, Model III
located along a 40-m transect line. The transect was positioned ANOVA (zone fixed factor, site random factor). Percentage
haphazardly and parallel to the shore, and the same line was used data were arcsine transformed, the density of juvenile corals was
for the census of juvenile corals and for the analysis of commu- square root-transformed, and all data were tested for the as-
nity structure (see below). Sea urchin size was defined as the sumptions of ANOVA with graphical analyses of the residuals.
maximum test diameter, and was measured to the nearest Statistical analyses were completed by using SYSTAT 5.2.
millimeter by using long-jawed calipers. The test diameters of the
first 100 Diadema encountered in randomly located, 1-m2 quad- Results
rats at each site were measured. Abundances of Diadema at each site were highest in a band
extending seaward from the shore (or reef crest) to form a
Benthic Community Sampling. Percent cover of major benthic contiguous zone 60 13 m in width (mean SE, pooled across
components, such as algal turf (in the sense of Carpenter; see ref. sites). Mean population densities of Diadema were more than an
17), macroalgae, crustose coralline algae, and live coral, were order of magnitude higher within the urchin zone ( 5 per m2
estimated in randomly located 0.25-m2 quadrats (n 20) in each with abundances reaching 12 per m2 in some locations) than in
zone. Quadrats were subdivided into 25 squares (each repre- the adjacent seaward algal zone at similar depths (Fig. 2), and
senting 4% of the quadrat), and the benthic component domi- differed among zones (F1,4 206.57, P 0.001). Diadema
nating each subdivision was recorded. densities did not vary significantly among sites, and the site–zone
interaction was not significant. In the areas between the five
Juvenile Coral Abundances and Sizes. Juvenile corals were defined study sites there were a few isolated pockets of Diadema that
as colonies between 2 mm (1 polyp) and 4 cm in diameter were surrounded by macroalgal-dominated substrata. Other
(20), and were counted by using randomly located 1-m2 quadrats species of echinoids were rare at all sites, and the highest mean
(n 10 per zone). Juvenile corals (i.e., not spat) were located abundances were 0.15 per m2 for Tripneustes ventricosus, 2.3 per
by carefully examining the substratum beneath the macroalgal m2 for Echinometra viridis, 0.24 per m2 for Lytechinus williamsi,
canopy when necessary and by removing sediment; they were and 0.15 per m2 for Eucidaris tribuloides.
identified to species or to genus when they lacked features The size–frequency distributions of Diadema in the sea urchin
allowing congeners to be distinguished. Siderastrea radians and zones were similar at all sites (data not shown), although mean
Favia fragum were omitted from all analyses because small sizes differed among sites (F4,190 2.75, P 0.03), with mean
colonies ( 4 cm diameter) of these species are sexually mature sizes varying 0.5 cm between sites. The size–frequency distri-
(21); inclusion of these species did not alter the patterns de- bution of Diadema pooled across sites is shown in Fig. 2 Inset.
scribed (data not shown). The sizes of juvenile corals were Benthic community structure differed greatly between the
measured in each zone by using calipers ( 0.1 mm), and size was zones where Diadema was abundant (urchin zone) and adjacent
5068 www.pnas.org cgi doi 10.1073 pnas.071524598 Edmunds and Carpenter
Fig. 3. Abundances of macroalgae and corals (Inset) in sea urchin and algal
zones at five sites along the north coast of Jamaica. (Bars mean percent
cover; error bars 1 SE.) See Fig. 1 for site designations; the order of sites in
the Inset is the same as for the main graph.
Fig. 4. Density and number of genera per quadrat (Inset) of juvenile corals
in sea urchin and algal zones at five sites along the north coast of Jamaica
ECOLOGY
areas where sea urchins were rare (algal zone). Percent cover of (mean SE; n 10 for each bar). See Fig. 1 for site designations; the order of
macroalgae was, on average, 10 times higher in areas where sites in the Inset is the same as for the main graph.
urchins were rare (Fig. 3), and differed significantly among zones
(F1,4 139.83, P 0.001). The pattern of higher macroalgal
pora cervicornis (n 3), Acropora palmata (n 1), Montastraea
abundance outside the urchin zone held across all sites, but was
annularis sensu lato (n 2), Montastraea cavernosa (n 2),
stronger at some sites than others, as demonstrated by the strong
Diploria spp. (n 11), and Colpophyllia natans (n 1). Juvenile
site–zone interaction (F4,190 6.58, P 0.001). The benthic
A. cervicornis, A. palmata, and M. annularis were found only in
community within the sea urchin zone was dominated by algal
the urchin zones. The sizes of all juvenile corals (pooled by
turfs (70 3% cover, mean SE) and crustose coralline algae
genus) were significantly different among sites (F4,859 6.08, P
(9 1% cover, mean SE). Abundances of algal turfs varied
0.001) but not between zones (F1,4 5.538, P 0.050), and the
across zones (F1,4 50.04, P 0.005) but not between sites, and
interaction was not significant. Mean sizes (pooled by genus and
there was a significant site–zone interaction (F4,190 12.94, P
zone) varied from 21.7 0.5 mm ( SE, n 180) at DB to 25.7
0.001). Percent cover of crustose coralline algae differed be- 0.6 mm ( SE, n 180) at LTS, with a grand mean size (pooled
tween sites (F4,190 14.76, P 0.001) but not between zones, and by genus, site, and zone) of 23.1 0.4 mm ( SE, n 745).
there was a significant site–zone interaction (F4,190 17.58, P Similarly, the sizes of juvenile Porites and Agaricia (the only two
0.001). The percent cover of live coral was 10% at four of five genera found at all five sites) varied significantly among sites
sites, and varied across sites (F4,190 19.76, P 0.001) but not (F4,220 5.20, P 0.010 and F4,286 3.30, P 0.050, respec-
between zones (Fig. 3), although a significant site–zone inter- tively) but not between zones, and there was no significant
action (F4,190 5.92, P 0.001) resulted from the high coral interaction. Thus, the size of juvenile corals was similar in urchin
coverage and large difference between zones at Dairy Bull. High and algal zones, and there were significant but relatively small
coral cover has been reported previously at the Dairy Bull differences ( 5 mm) among sites.
site (19).
In the sea urchin zone, juvenile corals were grazed around and Discussion
over by Diadema and were conspicuous against the substratum. The results of our surveys demonstrate that increased densities
Juvenile coral density, pooled by taxon, was between 2- and of sea urchins are associated with a reduction in cover by
11-fold higher in the sea urchin zone compared with the algal macroalgae and elevated densities of juvenile corals (up to
zone at all sites (Fig. 4). Overall, the density of juvenile corals 11-fold). Together with a plethora of correlational and experi-
was significantly higher in the sea urchin zone compared with the mental studies of Diadema grazing (8, 17), as well as recent
algal zone at all sites (F1,4 21.82, P 0.010); the mean value results based on 7 years of data at an adjacent site (12), our
in all five urchin zones was 24 per m2, with 43 per m2 at one findings suggest that Diadema caused the decline in macroalgae
site. There was a significant site–zone interaction (F4,90 4.42, and initiated a change in community structure at the spatial scale
P 0.010) as a result of the among-site differences in the examined. Given the rarity of Tripneustes on the shallow fore-
magnitude of the increase of juvenile-coral density in the sea reefs of Jamaica, the hypothesized synergistic role of this sea
urchin zone. urchin in initiating the removal of macroalgae (12, 16) is not
The juvenile corals at the five sites belonged to 14 genera and supported by the present results. Instead, our findings demon-
at least 17 species, and the number of genera per quadrat was up strate a putative phase reversal (from macroalgae) in the pres-
to 2.1 times higher in the sea urchin zone compared with the algal ence of high densities (up to 12 per m2) of Diadema and the
zone, although this difference was not significant (F1,4 10.20, near-complete absence of Tripneustes ( 0.15 per m2).
P 0.050, Fig. 4). The most common taxa of juvenile corals There are several mechanisms by which Diadema could have
encountered were Agaricia spp. (n 915), Porites spp. (n 167), facilitated the changes underway in Jamaica. The most conspic-
Siderastrea siderea (n 323), Leptoseris cucullata (n 46), and uous result of Diadema grazing is the removal of macroalgae and
Stephanocoenia michilini (n 40); there were small numbers of its replacement by a low biomass, high-turnover algal community
several primary reef-framework-building corals, including Acro- (algal turf) interspersed with crustose coralline algae (17).
Edmunds and Carpenter PNAS April 24, 2001 vol. 98 no. 9 5069
Experimental manipulations have demonstrated more subtle
effects, including the enhancement of recruitment and survivor-
ship of juvenile corals at intermediate densities of Diadema (4
per m2). While settlement of coral spat is highest in the absence
of Diadema, survivorship of juvenile corals is low due to algal
overgrowth. At high sea-urchin densities, intense grazing dam-
ages juvenile corals, and coral survivorship is reduced (22).
Additionally, increased cover of crustose coralline algae might
enhance coral settlement (23). Decreased survivorship of juve-
nile corals in the absence of grazing likely is caused by a variety
of factors including overgrowth, shading and or abrasion by
macroalgae (24), smothering by sediment that accumulates and
is stabilized by higher algal biomass (25), or a combination of
these effects. As a result, grazing by Diadema may increase the
abundance of juvenile corals through enhanced coral settlement,
and or increased survivorship of juveniles. Manipulative exper-
iments will be required to determine the relative importance of
these processes.
Further support for the pivotal role of Diadema in the recent
changes in Jamaica comes from the temporal coincidence of
events. Diadema densities increased at the same time that
macroalgal-free patches began to appear on the north coast of
Jamaica (in 1995–96; P.J.E., unpublished observations; ref. 12);
age estimates for Diadema and average-sized juvenile corals are
consistent with their recruitment close to, or after, 1995. For
Diadema, growth rates are rapid during their first year and slow
considerably as individuals approach a maximum size of 100
mm (test diameter) after 3–4 years (26). Growth rates and test
diameters also are affected by resource availability, often re- Fig. 5. (A) Population densities of Diadema at depths between 4 and 10 m
flected by the inverse relationship between Diadema population along the north coast of Jamaica from 1977 to 2000. Reported abundances are
density and mean individual size (27, 28). Given the abundant averaged over sites and depths within a sampling period (year) for each study.
algae on the study reefs in Jamaica and the low probability that (a) Discovery Bay (R.C.C., unpublished data), (b) Discovery Bay (36), (c) Discov-
Diadema growth was limited by resources, test diameter conse- ery Bay and Rio Bueno (37), (d) 14 sites in Jamaica (1), (e) Discovery Bay (12),
quently may allow estimation of age. For the range of maximum (f) algal zones at five sites (this study), (g) sea urchin zones at five sites (this
study). (B) Densities of juvenile corals along the north coast of Jamaica
test diameters measured in this study (10–89 mm), and assuming
between 1976 and 2000. There are no comparable data from the mid-1980s to
size-specific growth rates (29), the individuals range from ap- the early 1990s (d), but it is likely that there was little or no coral recruitment
proximately 1 month to 4 years old. The overall mean individual over this period (1). Densities of juvenile corals from the present study (aver-
size pooled over all sites is 56 mm with an estimated age of 2 aged across sites) for the algal and urchin zones (f and g, respectively). (h) The
years. For scleractinians, the relationship between age and size densities of juvenile corals ( 5 cm diameter) on Discovery Bay (11-m depth) in
is determined by the time-integrated exposure to partial mor- 1976 –1978 (34) calculated from ref. 38 (i) for foliaceous corals ( 50 cm2) at Rio
tality and growth rates. Older colonies likely have experienced Bueno (10-m depth) between 1977 and 1980, and calculated from ref. 39 (j) by
one or more partial mortality events and have a poor relationship using the densities of new recruits ( 2.6 cm diameter) of Agaricia agaricites
between size and age (30). Juvenile colonies are less likely to and Leptoseris cucullata at 10-m and 20-m depths, assuming that they repre-
sented 75% of the coral recruits. (k) The average densities of juvenile corals
have experienced partial mortality and, therefore, their size and
( 4 cm diameter) at four sites (10-m depth) on, or close to, Discovery Bay (19).
age are related more closely. The published growth rates of (l) Juvenile corals ( 4 cm diameter) at Dairy Bull (10-m depth; P.J.E., unpub-
juvenile corals are highly variable, ranging from 2 mm y (20) lished data).
to 36 mm y (31), but many grow at 12 mm y (31, 32). When
12 mm y is used as a working growth-rate estimate, juvenile
corals 23 mm in diameter (the mean size of all juveniles in this recruitment failure (Fig. 5). Significantly, the densities we report
study) are 23 months old, or about the same age as the here for both Diadema and juvenile corals are approaching those
average-sized Diadema. Whereas such calculations demonstrate recorded before the mass mortality of Diadema in 1983–84. In
that the concordance of ages of Diadema and juvenile corals is addition to the common genera of juvenile corals encountered
consistent with a cause-and-effect hypothesis, such correlational (e.g., Agaricia, Porites, and Siderastrea), all of which typically are
evidence cannot prove that Diadema has resulted in increased abundant recruits on Caribbean reefs (32), it also is ecologically
abundances of juvenile corals. Nevertheless, the recovery of significant that we found small numbers of recruits of Acropora
Diadema on the study reefs in Jamaica and its association with
cervicornis, A. palmata, and Montastraea annularis in the sea
elevated densities of juvenile corals (and reduced macroalgal
urchin zones, because these species are among the most impor-
cover) suggest that there is a functional relationship between
these events. Further studies are required to determine whether tant reef-framework-building corals in the Caribbean (33). Al-
coral cover in Jamaica will increase after decades of decline, or though the recruits of these species currently are found at low
whether the new state is stable. In January 2001, the sea urchin densities ( 0.4 per m2) along the north coast of Jamaica, this is
zones in Jamaica still contained 4.0 0.9 Diadema per m2 and routinely the case in studies of coral recruitment in the Carib-
were 60 2 m in width (both mean SE, n 5 sites), thereby bean (32). Moreover, the present-day densities are at least as
demonstrating persistence over at least 1 year. high as those recorded 24 years ago at Discovery Bay (34), when
When the present results are placed in a historical context, adult colonies of Acropora and Montastraea were the dominant
comparing abundances of Diadema and juvenile corals in Ja- species on Jamaican reefs (6, 14, 35).
maica over the last 20 years, it seems that the present abun- If the patterns documented here result in a reversal of the
dances are increasing after more than a decade of apparent phase shift from macroalgae to corals and algal turf on shallow
5070 www.pnas.org cgi doi 10.1073 pnas.071524598 Edmunds and Carpenter
reefs in Jamaica, it would indicate that macroalgal dominance of inevitable throughout the western Atlantic, this study does
Caribbean reefs is not an inevitable and terminal consequence provide good news about the recovery of highly degraded
of natural and anthropogenic disturbances. Instead, our results Caribbean coral reefs.
reemphasize the disproportionate effects of a single species
(Diadema) in mediating transitions between alternate states on We thank S. Genovese for facilitating the collaboration that generated this
present-day reefs, particularly those with reduced abundances of research, M. Haley for graciously hosting our stay at the Discovery Bay
Marine Laboratory, and R. Habeeb and C. Zilberberg for diving assistance.
herbivorous fishes (8). The coral reefs of Jamaica have been at This work was supported by the East West Marine Biology Program of
the forefront of reports of ecosystem collapse, and predictions of Northeastern University and, in part, by a National Institutes of Health
the future for most reefs remain gloomy (40, 41). Although our Grant Minority Biomedical Research Support GM48680 (to R.C.C.). This
results should not be construed to mean that reef recovery is is Discovery Bay Marine Laboratory contribution number 631.
1. Hughes, T. P. (1994) Science 265, 1547–1551. 19. Edmunds, P. J. & Bruno, J. F. (1996) Mar. Ecol. Prog. Ser. 143, 165–171.
2. Knowlton, N. (1992) Am. Zool. 32, 674–682. 20. Edmunds, P. J. (2000) Mar. Ecol. Prog. Ser. 202, 113–124.
3. Done, T. (1992) Hydrobiologia 247, 121–132. 21. Soong, K. (1993) Coral Reefs 12, 77–83.
4. Ostrander, G. K., Armstrong, K. M., Knobbe, E. T., Gerace, D. & Scully, E. P. 22. Sammarco, P. S. (1980) J. Exp. Mar. Biol. Ecol. 45, 245–272.
(2000) Proc. Natl. Acad. Sci. USA 97, 5297–5302. (First Published May 2, 2000; 23. Morse, D. E. N., Hooker, N., Morse, A. N. C. & Jensen, R. A. (1988) J. Exp.
10.1073 pnas.090104897) Mar. Biol. Ecol. 116, 193–217.
5. Aronson, R. B. & Precht, W. F. (2001) in Evolutionary Paleoecology: The 24. Miller, M. W. & Hay, M. E. (1996) Ecol. Monog. 66, 323–344.
Ecological Context of Macroevolutionary Change, eds. Allmon, W. D. & Bottjer, 25. Birkeland, C., Rowley, D. & Randall, R. H. (1981) Proc. 4th Int. Coral Reef
D. J. (Columbia Univ. Press, New York), 171–233. Symp. 2, 339–344.
6. Woodley, J. D., Chornesky, E. A., Clifford, P. A., Jackson, J. B. C., Kaufman, 26. Lewis, J. B. (1966) Bull. Mar. Sci. 16, 151–158.
L. S., Knowlton, N., Lang, J. C., Pearson, M. P., Porter, J. W., Rooney, M. C., 27. Carpenter, R. C. (1981) J. Mar. Res. 39, 749–765.
et al. (1981) Science 214, 749–755.
28. Levitan, D. R. (1988) Oecologia 76, 627–629.
7. Lapointe, B. E. (1997) Limnol. Oceanogr. 42, 1119–1131.
29. Levitan, D. R. (1991) Biol. Bull. 181, 261–268.
8. Lessios, H. R. (1988) Annu. Rev. Ecol. Syst. 19, 371–393.
30. Hughes, T. P. & Jackson, J. B. C. (1980) Science 209, 713–715.
9. Petraitis, P. S. & Laitham, R. E. (1999) Ecology 80, 429–442.
ECOLOGY
31. VanMoorsel, G. W. M. N. (1988) Mar. Ecol. Prog. Ser. 50, 127–135.
10. Hunter, C. L. & Evans, C. W. (1995) Bull. Mar. Sci. 57, 501–515.
32. Bak, R. P. M. & Engel, E. H. (1979) Mar. Biol. 54, 341–352.
11. Hughes, T. P., Szmant, A. M., Steneck, R. S., Carpenter, R. C. & Miller, S. L.
(1999) Limnol. Oceanogr. 44, 1583–1586. 33. Sheppard, C. R. C. (1982) Mar. Ecol. Prog. Ser. 7, 83–115.
12. Aronson, R. B. & Precht, W. F. (2000) Limnol. Oceanogr. 45, 251–255. 34. Rylaarsdam, K. W. (1983) Mar. Ecol. Prog. Ser. 13, 249–260.
13. Ginsburg, R. N., ed. (1997) Proceedings of the Colloquium of Global Aspects of 35. Liddell, W. D., Ohlhorst, S. L. & Boss, S. K. (1984) Paleontol. Am. 59, 385–389.
Coral Reefs: Health, Hazards and History (Rosenstiel School of Marine and 36. Morrison, D. (1988) Ecology 69, 1367–1382.
Atmospheric Science, University of Miami). 37. Hughes, T. P., Keller, B. D., Jackson, J. B. C. & Boyle, M. J. (1985) Bull. Mar.
14. Goreau, T. F. (1959) Ecology 40, 67–89. Sci. 36, 377–384.
15. Woodley, J. D. (1999) Coral Reefs 18, 192. 38. Hughes, T. P. & Jackson, J. B. C. (1985) Ecol. Monog. 55, 141–166.
16. Woodley, J. D., Gayle, P. M. H. & Judd, N. (1999) Coral Reefs 18, 193. 39. Hughes, T. P. (1985) Proc. 5th Int. Coral Reef Symp. 4, 101–106.
17. Carpenter, R. C. (1986) Ecol. Monogr. 56, 345–363. 40. Hoegh-Guldberg, O. (1999) Mar. Freshwater Res. 50, 839–866.
18. Liddell, W. D. & Ohlhorst, S. L. (1987) Bull. Mar. Sci. 40, 311–329. 41. Hughes, T. P. & Tanner, J. E. (2000) Ecology 81, 2250–2263.
Edmunds and Carpenter PNAS April 24, 2001 vol. 98 no. 9 5071
cover and increases abundance of juvenile
corals on a Caribbean reef
Peter J. Edmunds*† and Robert C. Carpenter†
Department of Biology, California State University, Northridge, CA 91330-8303
Edited by Robert T. Paine, University of Washington, Seattle, WA, and approved February 6, 2001 (received for review November 2, 2000)
The transition of many Caribbean reefs from coral to macroalgal opinion that reefs in the Caribbean now are in a highly degraded
dominance has been a prominent issue in coral reef ecology for state (13).
more than 20 years. Alternative stable state theory predicts that The coral reefs along the north coast of Jamaica, particularly
these changes are reversible but, to date, there is little indication at Discovery Bay, provide the most familiar example of the
of this having occurred. Here we present evidence of the initiation collapse of Caribbean reefs. Discovery Bay has been studied
of such a reversal in Jamaica, where shallow reefs at five sites along extensively since the 1950s (1, 5, 14), arguably more than any
8 km of coastline now are characterized by a sea urchin-grazed other reef in the Caribbean, and has remained at the forefront
zone with a mean width of 60 m. In comparison to the seaward of the reports of a macroalgal phase shift along the north coast
algal zone, macroalgae are rare in the urchin zone, where the of Jamaica. Virtually identical trends have been reported at
density of Diadema antillarum is 10 times higher and the density of multiple sites in Jamaica (1), but the majority of these sites have
juvenile corals is up to 11 times higher. These densities are close to been monitored less frequently than those at Discovery Bay. In
the 1950s, the reefs at Discovery Bay were characterized by small
ECOLOGY
those recorded in the late 1970s and early 1980s and are in striking
contrast to the decade-long recruitment failure for both Diadema amounts of macroalgae, and scleractinian corals covered as
and scleractinians. If these trends continue and expand spatially, much as 90% of the substratum (14). By the 1990s, after two
reefs throughout the Caribbean may again become dominated by major hurricanes, decades of overfishing, the near-complete loss
corals and algal turf. of the keystone echinoid Diadema (8), increasing human popu-
lation pressure, and possible nutrification, Jamaican reefs were
dominated by macroalgae to a depth of 35 m, and coral cover was
M any coral reefs throughout the Western Atlantic region
have undergone dramatic changes in community structure
over the past two decades. The best known examples of these
reduced to less than 5% (1). However, in 1992 there were small
patches of Diadema on the forereef of Discovery Bay (15), and
changes are found in the Caribbean (1), where reefs that were by 1995–96, Diadema had become locally abundant (e.g., 1.8 per
formerly dominated by scleractinian corals and diminutive algal m2, ref. 16) in shallow water ( 6-m depth) and formed larger
turfs have become overgrown by macroalgae. This transition is patches of macroalgal-free substratum (P.J.E., unpublished ob-
referred to often as a phase shift to an alternate state (1–4), servations). Typically, these patches were scalloped-shaped, tens
where the alternative state (i.e., one dominated by macroalgae) of meters in width, and contained locally dense populations of
is considered stable unless, or until, a reversal of one or more of the herbivorous echinoids Diadema and Tripneustes ventricosus
the causative agents favors another change. However, there are (12, 16). Tripneustes normally is rare or absent in forereef
habitats (16), and its co-occurrence with Diadema in 1995–96
few examples of multiple phase shifts on coral reefs (5) and none,
suggested that it might play a pivotal role in initiating the
to our knowledge, of a reversal from macroalgal to scleractinian
removal of macroalgae (12, 16). By January 2000, the expansion
dominance. Here we present evidence from the north coast of
and coalescence of macroalgal-free areas formed contiguous
Jamaica suggesting that such a change has been initiated. On
zones hundreds of meters in length, suggestive of a reversal in
these reefs, there is localized recovery of populations of the sea
community structure. This study was designed to quantify the sea
urchin Diadema antillarum, a decrease in the abundance of
urchin zones on a larger scale and test the prediction that
macroalgae, and an increase in the abundance of juvenile corals.
recovery of Diadema populations both enhances coral recruit-
Although there are few comparable data spanning the last 30
ment and facilitates a return to coral dominance.
years in Jamaica, there is evidence that the present densities of
Diadema and juvenile corals have not been reported for more Methods
than a decade. If these patterns persist, the widespread recovery Study Sites. Five sites were selected haphazardly along an 8-km
of Diadema populations alone may result in reefs dominated section of the north coast of Jamaica (Fig. 1). At each site, the
once again by scleractinian corals and algal turfs. sea urchin and algal zones sampled were parallel to the shore and
Phase shifts on Caribbean coral reefs can be caused by a to one another and were between 4.5 and 8.5 m in depth.
variety of anthropogenic and natural factors (1, 6–8), and are However, within a site, the sampled portions of the sea urchin
integral features of communities that demonstrate multiple and algal zones differed by 2 m or less in depth. At most sites,
stable states (2, 3). However, their detection is a function of the Diadema were found in depths as shallow as 1 m, and algal zones
spatiotemporal scale of investigation (2, 9), and evidence of extended to 20 m in depth. Widths of the zone of highest
phase shifts has come only from reefs such as those in Jamaica
(1) and Hawaii (10), where decades of data are available. Even
such long-term studies can provide only equivocal evidence of This paper was submitted directly (Track II) to the PNAS office.
the underlying mechanisms of change; thus, there is still debate See commentary on page 4822.
concerning the role of bottom-up [i.e., nutrification (7)] vs. *To whom reprint requests should be addressed. E-mail: peter.edmunds@csun.edu.
top-down [i.e., herbivore (11, 12)] control in mediating macroal- †P.J.E. and R.C.C. contributed equally to this work.
gal phase shifts. Regardless of the mechanism, the numerous The publication costs of this article were defrayed in part by page charge payment. This
reports of macroalgal phase shifts together with the absence of article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C.
any documented reversals have contributed to the popular §1734 solely to indicate this fact.
www.pnas.org cgi doi 10.1073 pnas.071524598 PNAS April 24, 2001 vol. 98 no. 9 5067–5071
Fig. 2. Population densities of Diadema antillarum in sea urchin and algal
zones at five sites along the north coast of Jamaica. (Bars mean population
densities; error bars 1 SE; see Fig. 1 for site designations.) The size–frequency
distribution of Diadema pooled across all sites (sea urchin zones only) is shown
in the Inset graph with the percent occurrence of nine size classes based on the
Fig. 1. Map of Jamaica showing the location of the study sites along the maximum test diameter. Size-class designations represent the following
north coast. RB Rio Bueno, LTS Long Term Survey, M1 Mooring 1, DB ranges of maximum test diameters: 1 20 mm, 2 20 –29 mm, 3 30 –39 mm,
Dairy Bull, EDB East Dairy Bull ( 0.7 km east of DB), and * Discovery Bay 4 40 – 49 mm, 5 50 –59 mm, 6 60 – 69 mm, 7 70 –79 mm, 8 80 – 89 mm,
Marine Laboratory. Sites were selected to sample reefs that have been the and 9 90 –99 mm.
subject of long-term studies [RB, LTS, M1 (refs. 1, 11, 15, and 18)] and recent
surveys (DB; ref. 19), and to span the greatest scale accessible with small boats
(e.g., EDB). taken as the average of the two major diameters of the basal
portion of each colony. The most common genera at each
site zone were measured (Leptoseris, Porites, Siderastrea, Steph-
abundance of Diadema were measured perpendicularly from anocoenia, and Agaricia), by using colonies (n 50, except for
shore (EDB, DB) or from the reefcrest (M1, LTS, RB) to the Agaricia at DB, where n 100) that were selected haphazardly
point where benthic community structure changed abruptly from while swimming along the transect line used for counting
algal turf dominance (i.e., Diadema-grazed, ref. 17) to macroal- juvenile corals and community structure.
gae (measurements at two randomly selected locations at each
site). All surveys were completed during January 2000. Statistical Analyses. Echinoid abundance, community structure
(i.e., percentage cover), density of juvenile corals, and the
Echinoid Abundances and Sizes. Abundances of sea urchins were number of genera of juvenile corals per quadrat were compared
estimated in 1-m2 quadrats (n 20 per zone) that were randomly among sites and between zones with a two-way, Model III
located along a 40-m transect line. The transect was positioned ANOVA (zone fixed factor, site random factor). Percentage
haphazardly and parallel to the shore, and the same line was used data were arcsine transformed, the density of juvenile corals was
for the census of juvenile corals and for the analysis of commu- square root-transformed, and all data were tested for the as-
nity structure (see below). Sea urchin size was defined as the sumptions of ANOVA with graphical analyses of the residuals.
maximum test diameter, and was measured to the nearest Statistical analyses were completed by using SYSTAT 5.2.
millimeter by using long-jawed calipers. The test diameters of the
first 100 Diadema encountered in randomly located, 1-m2 quad- Results
rats at each site were measured. Abundances of Diadema at each site were highest in a band
extending seaward from the shore (or reef crest) to form a
Benthic Community Sampling. Percent cover of major benthic contiguous zone 60 13 m in width (mean SE, pooled across
components, such as algal turf (in the sense of Carpenter; see ref. sites). Mean population densities of Diadema were more than an
17), macroalgae, crustose coralline algae, and live coral, were order of magnitude higher within the urchin zone ( 5 per m2
estimated in randomly located 0.25-m2 quadrats (n 20) in each with abundances reaching 12 per m2 in some locations) than in
zone. Quadrats were subdivided into 25 squares (each repre- the adjacent seaward algal zone at similar depths (Fig. 2), and
senting 4% of the quadrat), and the benthic component domi- differed among zones (F1,4 206.57, P 0.001). Diadema
nating each subdivision was recorded. densities did not vary significantly among sites, and the site–zone
interaction was not significant. In the areas between the five
Juvenile Coral Abundances and Sizes. Juvenile corals were defined study sites there were a few isolated pockets of Diadema that
as colonies between 2 mm (1 polyp) and 4 cm in diameter were surrounded by macroalgal-dominated substrata. Other
(20), and were counted by using randomly located 1-m2 quadrats species of echinoids were rare at all sites, and the highest mean
(n 10 per zone). Juvenile corals (i.e., not spat) were located abundances were 0.15 per m2 for Tripneustes ventricosus, 2.3 per
by carefully examining the substratum beneath the macroalgal m2 for Echinometra viridis, 0.24 per m2 for Lytechinus williamsi,
canopy when necessary and by removing sediment; they were and 0.15 per m2 for Eucidaris tribuloides.
identified to species or to genus when they lacked features The size–frequency distributions of Diadema in the sea urchin
allowing congeners to be distinguished. Siderastrea radians and zones were similar at all sites (data not shown), although mean
Favia fragum were omitted from all analyses because small sizes differed among sites (F4,190 2.75, P 0.03), with mean
colonies ( 4 cm diameter) of these species are sexually mature sizes varying 0.5 cm between sites. The size–frequency distri-
(21); inclusion of these species did not alter the patterns de- bution of Diadema pooled across sites is shown in Fig. 2 Inset.
scribed (data not shown). The sizes of juvenile corals were Benthic community structure differed greatly between the
measured in each zone by using calipers ( 0.1 mm), and size was zones where Diadema was abundant (urchin zone) and adjacent
5068 www.pnas.org cgi doi 10.1073 pnas.071524598 Edmunds and Carpenter
Fig. 3. Abundances of macroalgae and corals (Inset) in sea urchin and algal
zones at five sites along the north coast of Jamaica. (Bars mean percent
cover; error bars 1 SE.) See Fig. 1 for site designations; the order of sites in
the Inset is the same as for the main graph.
Fig. 4. Density and number of genera per quadrat (Inset) of juvenile corals
in sea urchin and algal zones at five sites along the north coast of Jamaica
ECOLOGY
areas where sea urchins were rare (algal zone). Percent cover of (mean SE; n 10 for each bar). See Fig. 1 for site designations; the order of
macroalgae was, on average, 10 times higher in areas where sites in the Inset is the same as for the main graph.
urchins were rare (Fig. 3), and differed significantly among zones
(F1,4 139.83, P 0.001). The pattern of higher macroalgal
pora cervicornis (n 3), Acropora palmata (n 1), Montastraea
abundance outside the urchin zone held across all sites, but was
annularis sensu lato (n 2), Montastraea cavernosa (n 2),
stronger at some sites than others, as demonstrated by the strong
Diploria spp. (n 11), and Colpophyllia natans (n 1). Juvenile
site–zone interaction (F4,190 6.58, P 0.001). The benthic
A. cervicornis, A. palmata, and M. annularis were found only in
community within the sea urchin zone was dominated by algal
the urchin zones. The sizes of all juvenile corals (pooled by
turfs (70 3% cover, mean SE) and crustose coralline algae
genus) were significantly different among sites (F4,859 6.08, P
(9 1% cover, mean SE). Abundances of algal turfs varied
0.001) but not between zones (F1,4 5.538, P 0.050), and the
across zones (F1,4 50.04, P 0.005) but not between sites, and
interaction was not significant. Mean sizes (pooled by genus and
there was a significant site–zone interaction (F4,190 12.94, P
zone) varied from 21.7 0.5 mm ( SE, n 180) at DB to 25.7
0.001). Percent cover of crustose coralline algae differed be- 0.6 mm ( SE, n 180) at LTS, with a grand mean size (pooled
tween sites (F4,190 14.76, P 0.001) but not between zones, and by genus, site, and zone) of 23.1 0.4 mm ( SE, n 745).
there was a significant site–zone interaction (F4,190 17.58, P Similarly, the sizes of juvenile Porites and Agaricia (the only two
0.001). The percent cover of live coral was 10% at four of five genera found at all five sites) varied significantly among sites
sites, and varied across sites (F4,190 19.76, P 0.001) but not (F4,220 5.20, P 0.010 and F4,286 3.30, P 0.050, respec-
between zones (Fig. 3), although a significant site–zone inter- tively) but not between zones, and there was no significant
action (F4,190 5.92, P 0.001) resulted from the high coral interaction. Thus, the size of juvenile corals was similar in urchin
coverage and large difference between zones at Dairy Bull. High and algal zones, and there were significant but relatively small
coral cover has been reported previously at the Dairy Bull differences ( 5 mm) among sites.
site (19).
In the sea urchin zone, juvenile corals were grazed around and Discussion
over by Diadema and were conspicuous against the substratum. The results of our surveys demonstrate that increased densities
Juvenile coral density, pooled by taxon, was between 2- and of sea urchins are associated with a reduction in cover by
11-fold higher in the sea urchin zone compared with the algal macroalgae and elevated densities of juvenile corals (up to
zone at all sites (Fig. 4). Overall, the density of juvenile corals 11-fold). Together with a plethora of correlational and experi-
was significantly higher in the sea urchin zone compared with the mental studies of Diadema grazing (8, 17), as well as recent
algal zone at all sites (F1,4 21.82, P 0.010); the mean value results based on 7 years of data at an adjacent site (12), our
in all five urchin zones was 24 per m2, with 43 per m2 at one findings suggest that Diadema caused the decline in macroalgae
site. There was a significant site–zone interaction (F4,90 4.42, and initiated a change in community structure at the spatial scale
P 0.010) as a result of the among-site differences in the examined. Given the rarity of Tripneustes on the shallow fore-
magnitude of the increase of juvenile-coral density in the sea reefs of Jamaica, the hypothesized synergistic role of this sea
urchin zone. urchin in initiating the removal of macroalgae (12, 16) is not
The juvenile corals at the five sites belonged to 14 genera and supported by the present results. Instead, our findings demon-
at least 17 species, and the number of genera per quadrat was up strate a putative phase reversal (from macroalgae) in the pres-
to 2.1 times higher in the sea urchin zone compared with the algal ence of high densities (up to 12 per m2) of Diadema and the
zone, although this difference was not significant (F1,4 10.20, near-complete absence of Tripneustes ( 0.15 per m2).
P 0.050, Fig. 4). The most common taxa of juvenile corals There are several mechanisms by which Diadema could have
encountered were Agaricia spp. (n 915), Porites spp. (n 167), facilitated the changes underway in Jamaica. The most conspic-
Siderastrea siderea (n 323), Leptoseris cucullata (n 46), and uous result of Diadema grazing is the removal of macroalgae and
Stephanocoenia michilini (n 40); there were small numbers of its replacement by a low biomass, high-turnover algal community
several primary reef-framework-building corals, including Acro- (algal turf) interspersed with crustose coralline algae (17).
Edmunds and Carpenter PNAS April 24, 2001 vol. 98 no. 9 5069
Experimental manipulations have demonstrated more subtle
effects, including the enhancement of recruitment and survivor-
ship of juvenile corals at intermediate densities of Diadema (4
per m2). While settlement of coral spat is highest in the absence
of Diadema, survivorship of juvenile corals is low due to algal
overgrowth. At high sea-urchin densities, intense grazing dam-
ages juvenile corals, and coral survivorship is reduced (22).
Additionally, increased cover of crustose coralline algae might
enhance coral settlement (23). Decreased survivorship of juve-
nile corals in the absence of grazing likely is caused by a variety
of factors including overgrowth, shading and or abrasion by
macroalgae (24), smothering by sediment that accumulates and
is stabilized by higher algal biomass (25), or a combination of
these effects. As a result, grazing by Diadema may increase the
abundance of juvenile corals through enhanced coral settlement,
and or increased survivorship of juveniles. Manipulative exper-
iments will be required to determine the relative importance of
these processes.
Further support for the pivotal role of Diadema in the recent
changes in Jamaica comes from the temporal coincidence of
events. Diadema densities increased at the same time that
macroalgal-free patches began to appear on the north coast of
Jamaica (in 1995–96; P.J.E., unpublished observations; ref. 12);
age estimates for Diadema and average-sized juvenile corals are
consistent with their recruitment close to, or after, 1995. For
Diadema, growth rates are rapid during their first year and slow
considerably as individuals approach a maximum size of 100
mm (test diameter) after 3–4 years (26). Growth rates and test
diameters also are affected by resource availability, often re- Fig. 5. (A) Population densities of Diadema at depths between 4 and 10 m
flected by the inverse relationship between Diadema population along the north coast of Jamaica from 1977 to 2000. Reported abundances are
density and mean individual size (27, 28). Given the abundant averaged over sites and depths within a sampling period (year) for each study.
algae on the study reefs in Jamaica and the low probability that (a) Discovery Bay (R.C.C., unpublished data), (b) Discovery Bay (36), (c) Discov-
Diadema growth was limited by resources, test diameter conse- ery Bay and Rio Bueno (37), (d) 14 sites in Jamaica (1), (e) Discovery Bay (12),
quently may allow estimation of age. For the range of maximum (f) algal zones at five sites (this study), (g) sea urchin zones at five sites (this
study). (B) Densities of juvenile corals along the north coast of Jamaica
test diameters measured in this study (10–89 mm), and assuming
between 1976 and 2000. There are no comparable data from the mid-1980s to
size-specific growth rates (29), the individuals range from ap- the early 1990s (d), but it is likely that there was little or no coral recruitment
proximately 1 month to 4 years old. The overall mean individual over this period (1). Densities of juvenile corals from the present study (aver-
size pooled over all sites is 56 mm with an estimated age of 2 aged across sites) for the algal and urchin zones (f and g, respectively). (h) The
years. For scleractinians, the relationship between age and size densities of juvenile corals ( 5 cm diameter) on Discovery Bay (11-m depth) in
is determined by the time-integrated exposure to partial mor- 1976 –1978 (34) calculated from ref. 38 (i) for foliaceous corals ( 50 cm2) at Rio
tality and growth rates. Older colonies likely have experienced Bueno (10-m depth) between 1977 and 1980, and calculated from ref. 39 (j) by
one or more partial mortality events and have a poor relationship using the densities of new recruits ( 2.6 cm diameter) of Agaricia agaricites
between size and age (30). Juvenile colonies are less likely to and Leptoseris cucullata at 10-m and 20-m depths, assuming that they repre-
sented 75% of the coral recruits. (k) The average densities of juvenile corals
have experienced partial mortality and, therefore, their size and
( 4 cm diameter) at four sites (10-m depth) on, or close to, Discovery Bay (19).
age are related more closely. The published growth rates of (l) Juvenile corals ( 4 cm diameter) at Dairy Bull (10-m depth; P.J.E., unpub-
juvenile corals are highly variable, ranging from 2 mm y (20) lished data).
to 36 mm y (31), but many grow at 12 mm y (31, 32). When
12 mm y is used as a working growth-rate estimate, juvenile
corals 23 mm in diameter (the mean size of all juveniles in this recruitment failure (Fig. 5). Significantly, the densities we report
study) are 23 months old, or about the same age as the here for both Diadema and juvenile corals are approaching those
average-sized Diadema. Whereas such calculations demonstrate recorded before the mass mortality of Diadema in 1983–84. In
that the concordance of ages of Diadema and juvenile corals is addition to the common genera of juvenile corals encountered
consistent with a cause-and-effect hypothesis, such correlational (e.g., Agaricia, Porites, and Siderastrea), all of which typically are
evidence cannot prove that Diadema has resulted in increased abundant recruits on Caribbean reefs (32), it also is ecologically
abundances of juvenile corals. Nevertheless, the recovery of significant that we found small numbers of recruits of Acropora
Diadema on the study reefs in Jamaica and its association with
cervicornis, A. palmata, and Montastraea annularis in the sea
elevated densities of juvenile corals (and reduced macroalgal
urchin zones, because these species are among the most impor-
cover) suggest that there is a functional relationship between
these events. Further studies are required to determine whether tant reef-framework-building corals in the Caribbean (33). Al-
coral cover in Jamaica will increase after decades of decline, or though the recruits of these species currently are found at low
whether the new state is stable. In January 2001, the sea urchin densities ( 0.4 per m2) along the north coast of Jamaica, this is
zones in Jamaica still contained 4.0 0.9 Diadema per m2 and routinely the case in studies of coral recruitment in the Carib-
were 60 2 m in width (both mean SE, n 5 sites), thereby bean (32). Moreover, the present-day densities are at least as
demonstrating persistence over at least 1 year. high as those recorded 24 years ago at Discovery Bay (34), when
When the present results are placed in a historical context, adult colonies of Acropora and Montastraea were the dominant
comparing abundances of Diadema and juvenile corals in Ja- species on Jamaican reefs (6, 14, 35).
maica over the last 20 years, it seems that the present abun- If the patterns documented here result in a reversal of the
dances are increasing after more than a decade of apparent phase shift from macroalgae to corals and algal turf on shallow
5070 www.pnas.org cgi doi 10.1073 pnas.071524598 Edmunds and Carpenter
reefs in Jamaica, it would indicate that macroalgal dominance of inevitable throughout the western Atlantic, this study does
Caribbean reefs is not an inevitable and terminal consequence provide good news about the recovery of highly degraded
of natural and anthropogenic disturbances. Instead, our results Caribbean coral reefs.
reemphasize the disproportionate effects of a single species
(Diadema) in mediating transitions between alternate states on We thank S. Genovese for facilitating the collaboration that generated this
present-day reefs, particularly those with reduced abundances of research, M. Haley for graciously hosting our stay at the Discovery Bay
Marine Laboratory, and R. Habeeb and C. Zilberberg for diving assistance.
herbivorous fishes (8). The coral reefs of Jamaica have been at This work was supported by the East West Marine Biology Program of
the forefront of reports of ecosystem collapse, and predictions of Northeastern University and, in part, by a National Institutes of Health
the future for most reefs remain gloomy (40, 41). Although our Grant Minority Biomedical Research Support GM48680 (to R.C.C.). This
results should not be construed to mean that reef recovery is is Discovery Bay Marine Laboratory contribution number 631.
1. Hughes, T. P. (1994) Science 265, 1547–1551. 19. Edmunds, P. J. & Bruno, J. F. (1996) Mar. Ecol. Prog. Ser. 143, 165–171.
2. Knowlton, N. (1992) Am. Zool. 32, 674–682. 20. Edmunds, P. J. (2000) Mar. Ecol. Prog. Ser. 202, 113–124.
3. Done, T. (1992) Hydrobiologia 247, 121–132. 21. Soong, K. (1993) Coral Reefs 12, 77–83.
4. Ostrander, G. K., Armstrong, K. M., Knobbe, E. T., Gerace, D. & Scully, E. P. 22. Sammarco, P. S. (1980) J. Exp. Mar. Biol. Ecol. 45, 245–272.
(2000) Proc. Natl. Acad. Sci. USA 97, 5297–5302. (First Published May 2, 2000; 23. Morse, D. E. N., Hooker, N., Morse, A. N. C. & Jensen, R. A. (1988) J. Exp.
10.1073 pnas.090104897) Mar. Biol. Ecol. 116, 193–217.
5. Aronson, R. B. & Precht, W. F. (2001) in Evolutionary Paleoecology: The 24. Miller, M. W. & Hay, M. E. (1996) Ecol. Monog. 66, 323–344.
Ecological Context of Macroevolutionary Change, eds. Allmon, W. D. & Bottjer, 25. Birkeland, C., Rowley, D. & Randall, R. H. (1981) Proc. 4th Int. Coral Reef
D. J. (Columbia Univ. Press, New York), 171–233. Symp. 2, 339–344.
6. Woodley, J. D., Chornesky, E. A., Clifford, P. A., Jackson, J. B. C., Kaufman, 26. Lewis, J. B. (1966) Bull. Mar. Sci. 16, 151–158.
L. S., Knowlton, N., Lang, J. C., Pearson, M. P., Porter, J. W., Rooney, M. C., 27. Carpenter, R. C. (1981) J. Mar. Res. 39, 749–765.
et al. (1981) Science 214, 749–755.
28. Levitan, D. R. (1988) Oecologia 76, 627–629.
7. Lapointe, B. E. (1997) Limnol. Oceanogr. 42, 1119–1131.
29. Levitan, D. R. (1991) Biol. Bull. 181, 261–268.
8. Lessios, H. R. (1988) Annu. Rev. Ecol. Syst. 19, 371–393.
30. Hughes, T. P. & Jackson, J. B. C. (1980) Science 209, 713–715.
9. Petraitis, P. S. & Laitham, R. E. (1999) Ecology 80, 429–442.
ECOLOGY
31. VanMoorsel, G. W. M. N. (1988) Mar. Ecol. Prog. Ser. 50, 127–135.
10. Hunter, C. L. & Evans, C. W. (1995) Bull. Mar. Sci. 57, 501–515.
32. Bak, R. P. M. & Engel, E. H. (1979) Mar. Biol. 54, 341–352.
11. Hughes, T. P., Szmant, A. M., Steneck, R. S., Carpenter, R. C. & Miller, S. L.
(1999) Limnol. Oceanogr. 44, 1583–1586. 33. Sheppard, C. R. C. (1982) Mar. Ecol. Prog. Ser. 7, 83–115.
12. Aronson, R. B. & Precht, W. F. (2000) Limnol. Oceanogr. 45, 251–255. 34. Rylaarsdam, K. W. (1983) Mar. Ecol. Prog. Ser. 13, 249–260.
13. Ginsburg, R. N., ed. (1997) Proceedings of the Colloquium of Global Aspects of 35. Liddell, W. D., Ohlhorst, S. L. & Boss, S. K. (1984) Paleontol. Am. 59, 385–389.
Coral Reefs: Health, Hazards and History (Rosenstiel School of Marine and 36. Morrison, D. (1988) Ecology 69, 1367–1382.
Atmospheric Science, University of Miami). 37. Hughes, T. P., Keller, B. D., Jackson, J. B. C. & Boyle, M. J. (1985) Bull. Mar.
14. Goreau, T. F. (1959) Ecology 40, 67–89. Sci. 36, 377–384.
15. Woodley, J. D. (1999) Coral Reefs 18, 192. 38. Hughes, T. P. & Jackson, J. B. C. (1985) Ecol. Monog. 55, 141–166.
16. Woodley, J. D., Gayle, P. M. H. & Judd, N. (1999) Coral Reefs 18, 193. 39. Hughes, T. P. (1985) Proc. 5th Int. Coral Reef Symp. 4, 101–106.
17. Carpenter, R. C. (1986) Ecol. Monogr. 56, 345–363. 40. Hoegh-Guldberg, O. (1999) Mar. Freshwater Res. 50, 839–866.
18. Liddell, W. D. & Ohlhorst, S. L. (1987) Bull. Mar. Sci. 40, 311–329. 41. Hughes, T. P. & Tanner, J. E. (2000) Ecology 81, 2250–2263.
Edmunds and Carpenter PNAS April 24, 2001 vol. 98 no. 9 5071